UCLA

Sensor technology plays an important role in modern information and intelligence. The accuracy of sensor measurement becomes more challenging in complex working environment. In this paper, we studied relationship between output frequency difference data and corresponding loading pressure in SAW (Surface Acoustic Wave) micro-pressure sensor. Then using frequency difference as input and pressure as output, we construct BP (Back Propagation) neural network which is trained using experimental data and used to predict output pressure of the sensor. We also calculate error with actual loading pressure, same in the least squares method commonly used. Through multiple comparisons of same set of sample data in overall and local accuracy of predicted results, we verified that the output error predicted by BP neural network is much smaller than least squares method. For example, one set of data is only about 2.9%. It provided a new method for data analysis in SAW micro-pressure sensor.